Deep learning-driven robust model predictive control based active cell equalisation for electric vehicle battery management system

Abstract

Stabilizing the cells in Electric Vehicle (EV) batteries allows for optimal efficiency, longer battery life, and greater performance. This research presents a deep learning tuning based on Robust Model Predictive Control (RMPC) to address the issue of EV cell imbalance. Deep learning is used to recognize the patterns of battery operation and hence the equalization of active cells is maintained. The equilibrium is maintained through the observation of the state of charge (SoC) of the cell. Parameters, such as Mean Absolute Error (MAE) and Mean Square Error (MSE) are employed to assess the efficiency of active cell balancing through the use of RMPC. The validity of proposed technique was shown by the use of MATLAB/Simulink in modelling, training, and testing the models as well as enhancing the battery performance. To perform the assessment, Multi-Layer Neural Network (MLNN), Long Short-Term Memory (LSTM) network and Recurrent Neural Network (RNN) are used. The proposed RMPC-based balancing demonstrated better accuracy with lower MSE and MAE values for RNN (0.712, 0.34), LSTM (0.724, 0.59), and MLNN (0.73, 0.65) as compared to Adaptive Model Predictive Control (AMPC) mechanism. The simulation results prove that proposed method efficiently provides the maximum voltage during the active cell balancing process.